Evidence is accumulating that early intervention can be effective in improving the skills of young children with autism spectrum disorder. However, the science is hampered by the lack of agreed "gold standard" tools for the measurement of progress and outcome. What is required is a reliable, valid, and sensitive measure of change in the core domains of autism, which can be undertaken blind to group and time. This article explores the use of a promising measure of change, for which reliability, validity, and sensitivity to change over a lengthy period have been previously demonstrated. Pilot data indicate that, despite some sensitivity to change over a short period of time, it does not capture treatment effects more effectively than an existing diagnostic tool. Future directions for the ongoing search are suggested, including consideration of how to achieve sensitivity to differential change as well as to change over time.